Pooler rollout update on 2 nodes is impossible with hard pod anti-affinity

• Which image of the operator are you using? ghcr.io/zalando/postgres-operator:v1.13.0
• Where do you run it - cloud or metal? Kubernetes or OpenShift? AWS EKS K8s
• Are you running Postgres Operator in production? yes
• Type of issue? feature request

I have 2 Kubernetes nodes for Postgres clusters (each cluster consists of 2 Postgres instances) and I need hard pod anti-affinity so the pods of the same Postgres cluster are always on different nodes.
But there is an issue that same pod anti-affinity and node-affinity settings apply to pooler pods.
So when starting update of pooler deployment, new pods just go into Pending state, because it have pooler pods on all nodes already.

I think it would be great to have a way to resolve it, at least possibility to specify different pod anti-affinity and node-affinity rules for pooler and Postgres pods.
Or maybe possibility to specify pooler deployment strategy to set maxSkew: 0 or type: Recreate.

Start by examining how the pooler deployment receives pod anti-affinity, node-affinity, and deployment strategy settings during an update. Reproduce the two-node rollout where new pooler pods remain Pending, then define and verify a configuration that lets pooler scheduling differ from Postgres pods or safely uses the requested rollout strategy.